First step towards implementing stack map descriptions for the optimizing compiler.

First step towards implementing stack map descriptions for the optimizing compiler.

Add a RawBitmap object and a bitmap builder.
Committed: https://code.google.com/p/dart/source/detail?r=5502

[srdjan, 2012-03-14 17:31:31]

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc
File vm/bitmap.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode28
vm/bitmap.cc:28: (size_ == kInitialSize) ? bit_list_ : bit_list_dynamic_;
You could instead have bit_list_dynamic_ point to bit_list_ and always refer to
bit_list_dynamic_ (could also rename them to bit_list_data_ and bit_list_).

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode41
vm/bitmap.cc:41: for (int32_t i = bound; i >= 0; i--) {
intptr_t

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode73
vm/bitmap.cc:73: }
Don't you need to clear the bits outside Minimum..Maximum range? Right now you
will clear only bits within the Minimum..Maximum.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h
File vm/bitmap.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode32
vm/bitmap.h:32: int32_t Maximum() const;
intptr_t instead of int32_t for the offsets? Here and elsewhere.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode50
vm/bitmap.h:50: int32_t BitSize() const { return (size_ * kBitsPerByte); }
Is it more readable to say SizeInBits?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode60
vm/bitmap.h:60: uint8_t bit_list_[kInitialSize];
Should this then be kInitialSizeInBytes? Size it typical word size, I think.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.cc
File vm/object.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.cc#newcode...
vm/object.cc:5395: if (Get(i)) return i;
Shouldn't this fail? i is Size() in first iteration and IsRange (offset <
Size()) should fail.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h
File vm/object.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2000
vm/object.h:2000: int32_t Maximum() const;
intptr_t, here and elsewhere

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2015
vm/object.h:2015: inline int32_t Size() const;
SizeInBits?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2022
vm/object.h:2022: void set_size(int32_t value) const;
SetSizeInBits?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.cc
File vm/raw_object.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.cc#new...
vm/raw_object.cc:142: intptr_t size = Smi::Value(raw_bitmap->ptr()->size_);
size in bits or words?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.h
File vm/raw_object.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.h#newc...
vm/raw_object.h:697: RawSmi* size_;  // Size of the bit map.
in bits or words?

[siva, 2012-03-14 22:54:20]

Addressed review comments.

I have renamed RawBitmap/Bitmap to RawStackmap/Stackmap to make it more clear in
the VM code.

PTAL.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc
File vm/bitmap.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode28
vm/bitmap.cc:28: (size_ == kInitialSize) ? bit_list_ : bit_list_dynamic_;
Good point. Done.

On 2012/03/14 17:31:32, srdjan wrote:
> You could instead have bit_list_dynamic_ point to bit_list_ and always refer
to
> bit_list_dynamic_ (could also rename them to bit_list_data_ and bit_list_).

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode41
vm/bitmap.cc:41: for (int32_t i = bound; i >= 0; i--) {
On 2012/03/14 17:31:32, srdjan wrote:
> intptr_t

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.cc#newcode73
vm/bitmap.cc:73: }
On 2012/03/14 17:31:32, srdjan wrote:
> Don't you need to clear the bits outside Minimum..Maximum range? Right now you
> will clear only bits within the Minimum..Maximum.

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h
File vm/bitmap.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode32
vm/bitmap.h:32: int32_t Maximum() const;
On 2012/03/14 17:31:32, srdjan wrote:
> intptr_t instead of int32_t for the offsets? Here and elsewhere.

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode50
vm/bitmap.h:50: int32_t BitSize() const { return (size_ * kBitsPerByte); }
On 2012/03/14 17:31:32, srdjan wrote:
> Is it more readable to say SizeInBits?

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/bitmap.h#newcode60
vm/bitmap.h:60: uint8_t bit_list_[kInitialSize];
On 2012/03/14 17:31:32, srdjan wrote:
> Should this then be kInitialSizeInBytes? Size it typical word size, I think.

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.cc
File vm/object.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.cc#newcode...
vm/object.cc:5395: if (Get(i)) return i;
Changed to (bound - 1).

Get(i) fails if we access out of bounds and hence that was also working.

On 2012/03/14 17:31:32, srdjan wrote:
> Shouldn't this fail? i is Size() in first iteration and IsRange (offset <
> Size()) should fail.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h
File vm/object.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2000
vm/object.h:2000: int32_t Maximum() const;
On 2012/03/14 17:31:32, srdjan wrote:
> intptr_t, here and elsewhere

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2015
vm/object.h:2015: inline int32_t Size() const;
On 2012/03/14 17:31:32, srdjan wrote:
> SizeInBits?

Done.

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/object.h#newcode2022
vm/object.h:2022: void set_size(int32_t value) const;
This size is not in bits it is just the size of the variable data area in
RawBitmap

On 2012/03/14 17:31:32, srdjan wrote:
> SetSizeInBits?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.cc
File vm/raw_object.cc (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.cc#new...
vm/raw_object.cc:142: intptr_t size = Smi::Value(raw_bitmap->ptr()->size_);
Renamed to size_in_bytes.

On 2012/03/14 17:31:32, srdjan wrote:
> size in bits or words?

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.h
File vm/raw_object.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3005/vm/raw_object.h#newc...
vm/raw_object.h:697: RawSmi* size_;  // Size of the bit map.
size in bytes, I have renamed it accordingly.

On 2012/03/14 17:31:32, srdjan wrote:
> in bits or words?

[srdjan, 2012-03-14 23:03:25]

LGTM

https://chromiumcodereview.appspot.com/9701010/diff/3010/vm/bitmap.h
File vm/bitmap.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3010/vm/bitmap.h#newcode14
vm/bitmap.h:14: class RawStackmap;
Alphabetical order

[siva, 2012-03-15 00:17:37]

Thanks.

https://chromiumcodereview.appspot.com/9701010/diff/3010/vm/bitmap.h
File vm/bitmap.h (right):

https://chromiumcodereview.appspot.com/9701010/diff/3010/vm/bitmap.h#newcode14
vm/bitmap.h:14: class RawStackmap;
On 2012/03/14 23:03:25, srdjan wrote:
> Alphabetical order

Done.